diff --git a/CHANGELOG.md b/CHANGELOG.md
index 07fb59bcd6c..eb6b9e62d2f 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -102,6 +102,7 @@ Docs: https://docs.openclaw.ai
 
 ### Fixes
 
+- Providers/xAI: stop sending OpenAI-style reasoning effort controls to native Grok Responses models, so `xai/grok-4.3` no longer fails live Docker/Gateway runs with `Invalid reasoning effort`.
 - Discord/gateway: measure heartbeat ACK timeouts from the actual heartbeat send, preventing late initial heartbeats from triggering false reconnect loops while the channel is still awaiting readiness. Fixes #77668. (#78087) Thanks @bryce-d-greybeard and @NikolaFC.
 - Discord/guilds: route plain text control commands such as `/steer` through the normal authorization and mention gate instead of silently dropping them before an agent session can see them. Fixes #78080. Thanks @ramitrkar-hash.
 - Control UI/Sessions: make the compaction count a compact `N Checkpoint(s)` disclosure and show expanded session-level details with modern checkpoint history cards across responsive table layouts. Thanks @BunsDev.
diff --git a/extensions/xai/index.test.ts b/extensions/xai/index.test.ts
index 347da642897..458e345cb53 100644
--- a/extensions/xai/index.test.ts
+++ b/extensions/xai/index.test.ts
@@ -221,7 +221,14 @@ describe("xai provider plugin", () => {
         model: createProviderModel({ id: "grok-4-1-fast" }),
       } as never),
     ).toMatchObject({
-      thinkingLevelMap: { off: null },
+      thinkingLevelMap: {
+        off: null,
+        minimal: null,
+        low: null,
+        medium: null,
+        high: null,
+        xhigh: null,
+      },
       compat: {
         toolSchemaProfile: "xai",
         nativeWebSearchTool: true,
diff --git a/extensions/xai/runtime-model-compat.ts b/extensions/xai/runtime-model-compat.ts
index c70b2cc57bf..ba2a4b11d82 100644
--- a/extensions/xai/runtime-model-compat.ts
+++ b/extensions/xai/runtime-model-compat.ts
@@ -7,13 +7,22 @@ type XaiRuntimeModelCompat = {
   >;
 };
 
+const XAI_UNSUPPORTED_REASONING_EFFORTS = {
+  off: null,
+  minimal: null,
+  low: null,
+  medium: null,
+  high: null,
+  xhigh: null,
+} satisfies NonNullable<XaiRuntimeModelCompat["thinkingLevelMap"]>;
+
 export function applyXaiRuntimeModelCompat<T extends XaiRuntimeModelCompat>(model: T): T {
   const withCompat = applyXaiModelCompat(model);
   return {
     ...withCompat,
     thinkingLevelMap: {
       ...withCompat.thinkingLevelMap,
-      off: null,
+      ...XAI_UNSUPPORTED_REASONING_EFFORTS,
     },
   };
 }
diff --git a/extensions/xai/stream.test.ts b/extensions/xai/stream.test.ts
index 520f800b229..115083a144a 100644
--- a/extensions/xai/stream.test.ts
+++ b/extensions/xai/stream.test.ts
@@ -1,6 +1,8 @@
 import type { StreamFn } from "@mariozechner/pi-agent-core";
 import type { Api, Context, Model } from "@mariozechner/pi-ai";
+import { streamSimpleOpenAIResponses } from "@mariozechner/pi-ai/openai-responses";
 import { describe, expect, it } from "vitest";
+import { applyXaiRuntimeModelCompat } from "./runtime-model-compat.js";
 import {
   createXaiFastModeWrapper,
   createXaiToolPayloadCompatibilityWrapper,
@@ -65,6 +67,44 @@ function runXaiToolPayloadWrapper(params: {
   );
 }
 
+async function captureXaiResponsesPayloadWithThinking(): Promise<Record<string, unknown>> {
+  const model = applyXaiRuntimeModelCompat({
+    api: "openai-responses",
+    provider: "xai",
+    id: "grok-4.3",
+    baseUrl: "https://api.x.ai/v1",
+    reasoning: true,
+    input: ["text", "image"],
+    cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
+    contextWindow: 1_000_000,
+    maxTokens: 64_000,
+  } as Model<"openai-responses">);
+
+  const payloadPromise = new Promise<Record<string, unknown>>((resolve, reject) => {
+    const timeout = setTimeout(
+      () => reject(new Error("provider payload callback was not invoked")),
+      1_000,
+    );
+    const stream = streamSimpleOpenAIResponses(
+      model,
+      { messages: [{ role: "user", content: "hello", timestamp: 0 }] },
+      {
+        apiKey: "test-api-key",
+        cacheRetention: "none",
+        reasoning: "low",
+        onPayload: (payload) => {
+          clearTimeout(timeout);
+          resolve(structuredClone(payload as Record<string, unknown>));
+          throw new Error("stop after payload capture");
+        },
+      },
+    );
+    void stream.result();
+  });
+
+  return await payloadPromise;
+}
+
 describe("xai stream wrappers", () => {
   it("rewrites supported Grok models to fast variants when fast mode is enabled", () => {
     expect(captureWrappedModelId({ modelId: "grok-3", fastMode: true })).toBe("grok-3-fast");
@@ -139,6 +179,13 @@ describe("xai stream wrappers", () => {
     expect(payload).not.toHaveProperty("reasoning_effort");
   });
 
+  it("marks native xAI Responses thinking efforts unsupported before pi-ai builds payloads", async () => {
+    const payload = await captureXaiResponsesPayloadWithThinking();
+
+    expect(payload).not.toHaveProperty("reasoning");
+    expect(payload).not.toHaveProperty("include");
+  });
+
   it("moves image-bearing tool results out of function_call_output payloads", () => {
     const payload: Record<string, unknown> = {
       input: [
diff --git a/extensions/xai/web-search.test.ts b/extensions/xai/web-search.test.ts
index 12503d57b3b..5ffec8f3431 100644
--- a/extensions/xai/web-search.test.ts
+++ b/extensions/xai/web-search.test.ts
@@ -585,7 +585,14 @@ describe("xai provider models", () => {
       api: "openai-responses",
       baseUrl: "https://api.x.ai/v1",
       reasoning: true,
-      thinkingLevelMap: { off: null },
+      thinkingLevelMap: {
+        off: null,
+        minimal: null,
+        low: null,
+        medium: null,
+        high: null,
+        xhigh: null,
+      },
       input: ["text", "image"],
       contextWindow: 1_000_000,
       maxTokens: 64_000,